Influence of the ocean surface temperature and sea ice concentration on regional climate changes in Eurasia in recent decades

Numerical experiments with the ECHAM5 atmospheric general circulation model have been performed in order to simulate the influence of changes in the ocean surface temperature (OST) and sea ice concentration (SIC) on climate characteristics in regions of Eurasia. The sensitivity of winter and summer climates to OST and SIC variations in 1998–2006 has been investigated and compared to those in 1968–1976. These two intervals correspond to the maximum and minimum of the Atlantic Long-Period Oscillation (ALO) index. Apart from the experiments on changes in the OST and SIC global fields, the experiments on OST anomalies only in the North Atlantic and SIC anomalies in the Arctic for the specified periods have been analyzed. It is established that temperature variations in Western Europe are explained by OST and SIC variations fairly well, whereas the warmings in Eastern Europe and Western Siberia, according to model experiments, are substantially (by a factor of 2–3) smaller than according to observational data. Winter changes in the temperature regime in continental regions are controlled mainly by atmospheric circulation anomalies. The model, on the whole, reproduces the empirical structure of changes in the winter field of surface pressure, in particular, the pressure decrease in the Caspian region; however, it substantially (approximately by three times) underestimates the range of changes. Summer temperature variations in the model are characterized by a higher statistical significance than winter ones. The analysis of the sensitivity of the climate in Western Europe to SIC variations alone in the Arctic is an important result of the experiments performed. It is established that the SIC decrease and a strong warming over the Barents Sea in the winter period leads to a cooling over vast regions of the northern part of Eurasia and increases the probability of anomalously cold January months by two times and more (for regions in Western Siberia). This effect is caused by the formation of the increased-pressure region with a center over the southern boundary of the Barents Sea during the SIC decrease and an anomalous advection of cold air masses from the northeast. This result indicates that, to estimate the ALO actions (as well as other long-scale climatic variability modes) on the climate of Eurasia, it is basically important to take into account (or correctly reproduce) Arctic sea ice changes in experiments with climatic models.     

Sea ice evolution over the 20th and 21st centuries as simulated by current AOGCMs

Outputs from simulations performed with current atmosphere-ocean general circulation models for the Fourth Assessment Report of Intergovernmental Panel on Climate Change (IPCC AR4) are used to investigate the evolution of sea ice over the 20th and 21st centuries. We first use the results from the “Climate of the 20th Century Experiment” to assess the ability of these models to reproduce the observed sea ice cover changes over the periods 1981–2000 and 1951–2000. The projected sea ice changes over the 21st century in response to the IPCC Special Report on Emission Scenarios A1B are then examined. Overall, there is a large uncertainty in simulating the present-day sea ice coverage and thickness and in predicting sea ice changes in both hemispheres. Over the period 1981–2000, we find that the multimodel average sea ice extent agrees reasonably well with observations in both hemipsheres despite the wide differences between the models. The largest uncertainties appear in the Southern Hemisphere. The climate change projections over the 21st century reveal that the annual mean sea ice extent decreases at similar rates in both hemispheres, and that the reduction in annual mean sea ice volume is about twice that of sea ice extent reduction in the Northern Hemisphere, in agreement with earlier studies. We show that the amplitude of the seasonal cycle of sea ice extent increases in both hemispheres in a warming climate, with a larger magnitude in the Northern Hemisphere. Furthermore, it appears that the seasonal cycle of ice extent is more affected than the one of ice volume. By the end of the 21st century, half of the model population displays an ice-free Arctic Ocean in late summer.     

北极各海域海冰覆盖范围的变化特征

Sea ice in the Arctic has been reducing rapidly in the past half century due to global warming.This study analyzes the variations of sea ice extent in the entire Arctic Ocean and its sub regions.The results indicate that sea ice extent reduction during 1979–2013 is most significant in summer,following by that in autumn,winter and spring.In years with rich sea ice,sea ice extent anomaly with seasonal cycle removed changes with a period of 4–6 years.The year of 2003–2006 is the ice-rich period with diverse regional difference in this century.In years with poor sea ice,sea ice margin retreats further north in the Arctic.Sea ice in the Fram Strait changes in an opposite way to that in the entire Arctic.Sea ice coverage index in melting-freezing period is an critical indicator for sea ice changes,which shows an coincident change in the Arctic and sub regions.Since 2002,Region C2 in north of the Pacific sector contributes most to sea ice changes in the central Aarctic,followed by C1 and C3.Sea ice changes in different regions show three relationships.The correlation coefficient between sea ice coverage index of the Chukchi Sea and that of the East Siberian Sea is high,suggesting good consistency of ice variation.In the Atlantic sector,sea ice changes are coincided with each other between the Kara Sea and the Barents Sea as a result of warm inflow into the Kara Sea from the Barents Sea.Sea ice changes in the central Arctic are affected by surrounding seas.     

北极秋季海冰减少与亚洲大陆冬季温度异常

本文使用SVD等诊断分析方法探讨北极秋季海冰密集度与亚洲冬季温度异常之间的关系。结果表明,近30余年来,北极秋季海冰减少伴随着亚洲大陆冬季温度降低,但青藏高原地区、北冰洋和北太平洋沿岸除外。北极秋季海冰密集度减小激发欧亚大陆和北冰洋北部两个区域位势高度的改变,这种异常的变化模态从秋季持续到冬季。位势高度异常的负值中心位于巴伦支海和喀拉海。位势高度异常的正值中心位于蒙古区域。与重力位势高度异常伴随的风场异常为亚洲冬季温度降低提供自北向南的冷气流。随着北极海冰的不断减少,其与亚洲大陆冬季温度降低之间的关系将为气候长期预测提供参考。     

全球热带海洋海表温度场异常对北极海冰的影响

本文利用1951?2021年哈德莱中心提供的海冰和海温最新资料以及美国国家海洋和大气管理局气候预报中心提供的NCEP/NCAR再分析资料,分析探讨了北极海冰70余年的长期变化特征,进而研究了其快速减少与热带海温场异常变化之间的联系,揭示了在全球热带海洋海温场变化与北极海冰之间存在密切联系的事实。结果表明,北极海冰异常变化最显著区域出现在格陵兰海、卡拉海和巴伦支海。热带不同海区对北极海冰的影响存在明显时滞时间和强度差异,热带大西洋的影响相比偏早,印度洋次之,太平洋偏晚。热带大西洋、印度洋和中东太平洋海温异常影响北极海冰的最佳时间分别是后者滞后26个月、30个月和34个月,全球热带海洋影响北极海冰的时滞时间为33个月。印度洋SST对北极海冰的影响程度最强,其次是太平洋,最弱是大西洋。全球热带海洋对北极海冰的影响过程中,热带东太平洋和印度洋起主导作用。当全球热带海洋SST出现正（负）距平时,北极海冰会出现偏少（多）的趋势,而AO、PNA、NAO对北极海冰变化起重要作用,是热带海洋与北极海冰相系数的重要“纽带”。而AO、PNA和NAO不仅受热带海洋SST的影响,同时也受太平洋年代际振荡PDO和大西洋多年代际AMO的影响,这一研究为未来北极海冰快速减少和全球气候变暖机理的深入研究提供理论支撑。     

Simulation of climate changes in the 20th–22nd centuries with a coupled atmosphere-ocean general circulation model

The results of numerical experiments with a coupled atmosphere-ocean general circulation model on the reproduction of climate changes during the 20th century and on the simulation of possible climate changes during the 21st–22nd centuries according to three IPCC scenarios of variations in the concentrations of greenhouse and other gases, as well as the results of the experiments with the doubled and quadruple concentrations of CO₂, are considered. An increase in the near-surface air temperature during the 20th century and the features of the observed climate changes, such as warming in 1940–1950 and its slowing down in 1960–1970, are adequately reproduced in the model. According to the model, the air-temperature increase during the 22nd century (as compared to the end of the 20th century) varies from 2 K for the most moderate scenario to 5 K for the warmest scenario. This estimate is somewhat lower than the expected warming averaged over the data of all models presented in the third IPCC report. According to model data, in the 22nd century, under all scenarios, at the end of summer, a complete or almost complete sea-ice melting will occur in the Arctic. According to the model, by the year 2200, the sea level will vary by 20 to 45 cm as compared to the level at the end of the 20th century.     

北极海冰的年代际转型与中国冻雨年代际变化的关系

基于1961-2013年HadISST海冰密集度资料,定义了北极海冰的季节性融冰指数,结果显示近几十年来北极季节性融冰范围呈显著的上升趋势,并分别在20世纪70年代末和90年代中期存在显著的年代际转型,相应地,中国冻雨发生频数总体上呈现出显著的减少趋势,但也存在显著的年代际转型。在20世纪70年代末之前,北极季节性融冰范围较小但显著增长,中国冻雨频数年际变化振幅较大,且主要受巴伦支海、喀拉海海冰的影响;20世纪70年代末至90年代中期北极季节性融冰范围维持振荡特征,没有显著的线性趋势,中国冻雨频数变化振幅减小,与北极海冰相关较弱,主要相关因子为北大西洋及北太平洋海表温度变化;而90年代中期以后,北极海冰融化加快,特别是2007年以后,季节性融冰范围大大增加,而中国冻雨频数处于低发时段,其变化与太平洋扇区海冰及堪察加半岛附近海温呈显著负相关,季节性融冰的显著区域也从东西伯利亚海逆时针旋转向波弗特海-加拿大群岛北部扩张,同时向北极中央区扩张。不同年代影响冻雨的海温或海冰关键海区不同,产生特定的大气环流异常响应,进而影响到我国冻雨。     

利用综合评分方法评估CMIP5模式对北极海冰的模拟能力

北极海冰冰盖自20世纪以来经历了前所未有的缩减，这使得北极海冰异常对大气环流的反馈作用日益显现。尽管目前的气候模式模拟北极海冰均为减少的趋势，但各模式间仍然存在较大的分散性。为了评估模式对于北极海冰变化及其气候效应的模拟能力，我们将海冰线性趋势和年际异常两者结合起来构造了一种合理的衡量指标。我们还强调巴伦支与卡拉海的重要性，因为前人研究证明此区域海冰异常是近年来影响大尺度大气环流变异的关键因子。根据我们设定的标准，CMIP5模式对海冰的模拟可被归为三种类型。这三组多模式集合平均之间存在巨大的差异，验证了这种分组方法的合理性。此外，我们还进一步探讨了造成模式海冰模拟能力差别的潜在物理因子。结果表明模式所采用的臭氧资料集对海冰模拟能力有显著的影响。     

Arctic salinity anomalies and their link to the North Atlantic during a positive phase of the Arctic Oscillation

Many of the changes observed during the last two decades in the Arctic Ocean and adjacent seas have been linked to the concomitant abrupt decrease of the sea level pressure in the central Arctic at the end of the 1980s. The decrease was associated with a shift of the Arctic Oscillation (AO) to a positive phase, which persisted throughout the mid 1990s. The Arctic salinity distribution is expected to respond to these dramatic changes via modifications in the ocean circulation and in the fresh water storage and transport by sea ice. The present study investigates these different contributions in the context of idealized ice-ocean experiments forced by atmospheric surface wind-stress or temperature anomalies representative of a positive AO index.Wind stress anomalies representative of a positive AO index generate a decrease of the fresh water content of the upper Arctic Ocean, which is mainly concentrated in the eastern Arctic with almost no compensation from the western Arctic. Sea ice contributes to about two-third of this salinification, another third being provided by an increased supply of salt by the Atlantic inflow and increased fresh water export through the Canadian Archipelago and Fram Strait. The signature of a saltier Atlantic Current in the Norwegian Sea is not found further north in both the Barents Sea and the Fram Strait branches of the Atlantic inflow where instead a widespread freshening is observed. The latter is the result of import of fresh anomalies from the subpolar North Atlantic through the Iceland-Scotland Passage and enhanced advection of low salinity waters via the East Icelandic Current. The volume of ice exported through Fram Strait increases by 20% primarily due to thicker ice advected into the strait from the northern Greenland sector, the increase of ice drift velocities having comparatively less influence. The export anomaly is comparable to those observed during events of Great Salinity Anomalies and induces substantial freshening in the Greenland Sea, which in turn contributes to increasing the fresh water export to the North Atlantic via Denmark Strait. With a fresh water export anomaly of 7 mSv, the latter is the main fresh water supplier to the subpolar North Atlantic, the Canadian Archipelago contributing to 4.4 mSv.The removal of fresh water by sea ice under a positive winter AO index mainly occurs through enhanced thin ice growth in the eastern Arctic. Winter SAT anomalies have little impact on the thermodynamic sea ice response, which is rather dictated by wind driven ice deformation changes. The global sea ice mass balance of the western Arctic indicates almost no net sea ice melt due to competing seasonal thermodynamic processes. The surface freshening and likely enhanced sea ice melt observed in the western Arctic during the 1990s should therefore be attributed to extra-winter atmospheric effects, such as the noticeable recent spring-summer warming in the Canada-Alaska sector, or to other modes of atmospheric circulations than the AO, especially in relation to the North Pacific variability.     

Structure of temperature variability in the high latitudes of the Northern Hemisphere

The spatial structure of surface air temperature (SAT) anomalies in the extratropical latitudes of the Northern Hemisphere (NH) during the 20th century is studied from the data obtained over the period 1892–1999. The expansion of the mean (over the winter and summer periods) SAT anomalies into empirical orthogonal functions (EOFs) is used for analysis. It is shown that variations in the mean air temperature in the Arctic region (within the latitudes 60°–90°N) during both the winter and summer periods can be described with a high accuracy by two spatial orthogonal modes of variability. For the winter period, these are the EOF related to the leading mode of variability of large-scale atmospheric circulation in the NH, the North Atlantic Oscillation, and the spatially localized (in the Arctic) EOF, which describes the Arctic warming of the mid-20th century. The expansion coefficient of this EOF does not correlate with the indices of atmospheric circulation and is hypothetically related to variations in the area of the Arctic ice cover that are due to long-period variations in the influx of oceanic heat from the Atlantic. On the whole, a significantly weaker relation to the atmospheric circulation is characteristic of the summer period. The first leading variability mode describes a positive temperature trend of the past decades, which is hypothetically related to global warming, while the second leading EOF describes a long-period oscillation. On the whole, the results of analysis suggest a significant effect of natural climatic variability on air-temperature anomalies in the NH high latitudes and possible difficulties in isolating an anthropogenic component of climate changes.     

Analysis of the Arctic sea ice conditions for 2011 at the onset of summer minimum

The summer minimum of the Arctic sea ice area and extent have been estimated for 2011 using satellite passive microwave data. Compared with sea ice conditions during the satellite era (1979 to the present), the Arctic ice cap is close in size to the absolute minimum recorded in 2007. However, the spatial distribution of sea ice at the end of summer differed in 2007 and 2011 due to the atmospheric circulation effect on the position of the ice edge. It is shown that the decreasing rate of the ice cover has increased fourfold since 2003. A linear model has been developed for the global short-term prediction of Arctic ice conditions and historical reconstruction (until the middle of the 20th century, including the pre-satellite era) of summer ice conditions from the air temperature fields using the dimensionality reduction technique (principal component analysis) and canonical correlation analysis. The simulation results confirm the drastic change in the sea ice area at the end of summer after 2002.     

北极冬季季节性海冰双模态特征分析

近年来北极海冰快速变化,北极中央区边缘正由以多年冰为主转为季节性海冰为主。通过对北极冬季季节性海冰的EOF分解发现,2002-2012年期间北极季节性海冰变化的前两模态主要体现为2005年和2007年的季节性海冰距平。其中第二模态主要体现了北极海冰在2005年的一种极端变化,而第一模态不仅体现了北极海冰在2007年的变化,还体现了北极季节性海冰的从负位相到正位相的转变。通过比较发现,在研究时段北极季节性海冰最主要的变化发生在北极太平洋扇区,在2007年,冬季季节性海冰距平发生位相转变,2007-2010年一直维持正位相,北极太平洋扇区冬季季节性海冰保持显著正距平。太平洋扇区表面温度最大异常也发生在2007年,从大气环流来看,2007年之后波弗特海区异常高压有利于夏季太平洋扇区海冰的减少,而西风急流的减弱有利于夏季波弗特海区异常高压的维持,结合夏季海冰速度,顺时针的冰速分布有利于海冰离开太平洋扇区,因而会导致冬季太平洋扇区季节性海冰转为正距平并且从2007年一直维持到2010年。     

地球系统模式FIO-ESM对北极海冰的模拟和预估

评估了地球系统模式FIO-ESM(First Institute of Oceanography-Earth System Model)基于CMIP5(Coupled Model Intercomparison Project Phase 5)的历史实验对北极海冰的模拟能力,分析了该模式基于CMIP5未来情景实验在不同典型浓度路径(RCPs,Representative Concentration Pathways)下对北极海冰的预估情况。通过与卫星观测的海冰覆盖范围资料相比,该模式能够很好地模拟出多年平均海冰覆盖范围的季节变化特征,模拟的气候态月平均海冰覆盖范围均在卫星观测值±15%范围以内。FIO-ESM能够较好地模拟1979-2005年期间北极海冰的衰减趋势,模拟衰减速度为每年减少2.24×104 km2,但仍小于观测衰减速度(每年减少4.72×104 km2)。特别值得注意的是:不同于其他模式所预估的海冰一直衰减,FIO-ESM对21世纪北极海冰预估在不同情景下呈现不同的变化趋势,在RCP2.6和RCP4.5情景下,北极海冰总体呈增加趋势,在RCP6情景下,北极海冰基本维持不变,而在RCP8.5情景下,北极海冰呈现继续衰减趋势。     

The Effect of Seasonal Variability of Atlantic Water on the Arctic Sea Ice Cover

Under the influence of global warming, the sea ice in the Arctic Ocean (AO) is expected to reduce with a transition toward a seasonal ice cover by the end of this century. A comparison of climate-model predictions with measurements shows that the actual rate of ice cover decay in the AO is higher than the predicted one. This paper argues that the rapid shrinking of the Arctic summer ice cover is due to its increased seasonality, while seasonal oscillations of the Atlantic origin water temperature create favorable conditions for the formation of negative anomalies in the ice-cover area in winter. The basis for this hypothesis is the fundamental possibility of the activation of positive feedback provided by a specific feature of the seasonal cycle of the inflowing Atlantic origin water and the peaking of temperature in the Nansen Basin in midwinter. The recently accelerated reduction in the summer ice cover in the AO leads to an increased accumulation of heat in the upper ocean layer during the summer season. The extra heat content of the upper ocean layer favors prerequisite conditions for winter thermohaline convection and the transfer of heat from the Atlantic water (AW) layer to the ice cover. This, in turn, contributes to further ice thinning and a decrease in ice concentration, accelerated melting in summer, and a greater accumulation of heat in the ocean by the end of the following summer. An important role is played by the seasonal variability of the temperature of AW, which forms on the border between the North European and Arctic basins. The phase of seasonal oscillation changes while the AW is moving through the Nansen Basin. As a result, the timing of temperature peak shifts from summer to winter, additionally contributing to enhanced ice melting in winter. The formulated theoretical concept is substantiated by a simplified mathematical model and comparison with observations.     

盛夏北极边缘海冰与来年冬季NAO之联系

许多研究认为,只有北大西洋涛动(NAO)是一种具有物理意义的模态,而北极涛动(AO)则是EOF分解得到的一种统计假象模态。为了从一个新的角度进一步探讨二者的差别,我们运用附条件的最大协方差分析(CMCA)统计了前期北极边缘海冰密集度(MSCI)与来年冬季NAO之间的跨季节遥相关关系,其中的ENSO信号和线性趋势已经在分析之前被去除。统计显著性结果表明:冬季负位相的NAO信号可以追溯到6个月前自盛夏开始至早冬季节北极MSCI异常的逐步演变。然而根据先前的研究,北极海冰异常仅可以超前冬季AO 大概4个月表现出显著信号。这表明盛夏北极MSCI的持续异常对来年冬季NAO的影响比对AO更强,同时也从另一个角度证实了AO与NAO确实存在差异。进一步分析还表明,前期MSCI异常的逐步演变主要与海表面热通量及气温异常有关。此外,我们还重新审视了负位相的NAO对北半球冬季气候异常的影响以及可能的物理机制。     

海冰消融背景下北极增温的季节差异及其原因探讨

运用哈德莱中心第一套海冰覆盖率(HadISST1)、欧洲中心(ERA_Interim)的温度以及NCEP第一套地表感热通量、潜热通量等资料,研究了1979—2011年33a来北极海冰消融的季节特点和空间特征,并从反照率——温度正反馈与地表感热通量、潜热通量等方面分析了海冰减少对北极增温影响的季节差异。结果表明,北极海冰在秋季和夏季的减少范围明显大于冬季和春季,而北极地表升温却在秋季和冬季最显著,夏季最为微弱,且夏季的增温趋势廓线也与秋冬季显著不同。这主要是因为夏季是融冰季,海冰融化将吸收潜热。且此时北极低空大气温度高于海表温度,海水相当于大气的冷源。随着海冰的消融,更多的热量由大气传入海洋用于融冰和加热上层海水,这使得夏季的低空大气不能显著升温。而在秋冬季,海冰凝结释放潜热,且此时低空大气温度远低于海水温度,海冰的减少使得海水将更多热量释放到大气中导致低空大气显著增暖。海水对大气的这种延迟放热机制是北极低空在夏季增温不显著而在秋冬季增温显著的主要原因。此外,秋冬季的海冰减少与北极近地面升温具有非常一致的空间分布,北冰洋东南边缘和巴伦支海北部分别是秋季和冬季海气相互作用的关键区域。     

基于CryoSat-2数据的2010-2017年北极海冰厚度和体积的季节与年际变化特征

北极海冰变化影响着全球物质平衡、能量交换和气候变化。本文基于CryoSat-2测高数据和OSI SAF海冰密集度及海冰类型产品，分析了2010-2017年北极海冰面积、厚度和体积的季节和年际变化特征，结合NCEP再分析资料探讨了融冰期北极气温异常和夏季风异常对海冰变化的影响。结果表明，结冰期海冰面积的增加量波动较大，海冰厚度的增加量呈明显下降趋势。融冰期海冰厚度的减小量波动较大，2013年以后融冰期海冰面积的减小量逐年增加。海冰体积的变化趋势和面积变化更相似，融冰期的减小速率大于结冰期的增加速率。融冰期北极海表面大气温度异常与海冰融化量正相关。夏季风影响海冰的辐合和辐散，在弗拉姆海峡海冰的输运过程中起关键作用，促进了北冰洋表层水向大洋深层的传输。     

The effect of tides on the volume of sea ice in the Arctic Ocean

Tides are believed to drive vertical mixing in the Arctic Ocean, thereby helping heat to reach the bottom of the sea ice layer, especially in regions with thick ice covers. However, tides are usually not included in ocean models. We investigated the effect of tides on sea ice in the Arctic Ocean using an ice-coupled ocean model that includes tides simultaneously. We found that with tidal forcing, the volume of sea ice increased by 8.5% in Baffin Bay, whereas it decreased by 17.8% in the Canadian Arctic Archipelago. The increase in sea ice volume in Baffin Bay results from the convergence of sea ice, driven by tidal residual currents. In contrast, the decrease in ice volume in the Canadian Archipelago is due to the suppression of ice formation in winter, especially in areas with steep topography, where the vertical mixing of temperature is enhanced by tides. Our results imply that tides should be directly included into the oceanic general circulation model (OGCM) to realistically reproduce the distribution of sea ice in the Arctic Ocean.     

The Iceland–Lofotes pressure difference: different states of the North Atlantic low-pressure zone

The extended North Atlantic low-pressure zone exhibits two pressure minima in the long-term winter mean: the primary one west of Iceland and the secondary one near Norwegian Lofotes Islands. Based on the ERA-40 data set and on wintertime monthly sea level pressure (SLP) anomalies at both places, the states of co- and antivariability are investigated. The covariability represents states of a strongly or weakly developed North Atlantic low-pressure zone. The difference between these two states represents the NAO pattern. The antivariability is defined by an Iceland–Lofotes difference (ILD) index, which is positive (negative) when the anomaly in the Lofotes area is higher (lower) than that in the Iceland area. An ILD pattern is calculated as difference between SLP composites for high and low ILD indices. The ILD pattern extends horizontally beyond the two centers and affects other prominent Northern Hemisphere pressure centres: Aleutian low; Siberian high and Azores high. The pattern extends into the stratosphere and shows significant impacts on surface air temperature, Arctic sea ice concentration and sea ice motion.     

Quantifying the contribution of natural variability to September Arctic sea ice decline

Arctic sea ice extent has been declining in recent decades. There is ongoing debate on the contribution of natural internal variability to recent and future Arctic sea ice changes. In this study, we contrast the trends in the forced and unforced simulations of carefully selected global climate models with the extended observed Arctic sea ice records. The results suggest that the natural variability explains no more than 42.3% of the observed September sea ice extent trend during 35 a(1979–2013) satellite observations, which is comparable to the results of the observed sea ice record extended back to 1953(61 a, less than 48.5% natural variability). This reinforces the evidence that anthropogenic forcing plays a substantial role in the observed decline of September Arctic sea ice in recent decades. The magnitude of both positive and negative trends induced by the natural variability in the unforced simulations is slightly enlarged in the context of increasing greenhouse gases in the 21st century.However, the ratio between the realizations of positive and negative trends change has remained steady, which enforces the standpoint that external forcing will remain the principal determiner of the decreasing Arctic sea ice extent trend in the future.